Abstract
Big data has become a trend in our world today. The rise in data production
rate, which has as consequence the increase in data quantity, comes with many
important challenges, one of which being data processing. The processing of big data
stands as the door that blocks the many benefits of big data on the unknown side.
Current data processing methods are found to be wanting when it comes to big data
processing, introducing, therefore, the need for new processing methods to handle big
data. Processing methods for big data can be built from scratch, that is, based on new
paradigms and approaches, or they can be created from the evolution of existing
methods already well known.
The study conducted in the present work sought to investigate the coupling of two or
more already existing approaches or tools from parallel disciplines, an evolution
approach, as a viable route for the development of a big data processing method so to
make big data benefits a reality. In order to achieve this objective, this study
conducted a solid literature review on traditional learning models and big data.
Furthermore, this study conducted a minimalistic experimentation in order to compare
the processing speed of the selected traditional learning models with and without big
data tools, as well as, comparing selected performance metrics values of traditional
learning models with and without a big data tool.
A faster processing time was recorded when traditional learning models was
implemented with the big data tools. However, the performance metrics were
relatively unchanged. A faster processing time implied that traditional learning model
used together with big data tools solved the need to address the volume issue of big
data analytics. Unchanged performance metrics signify that big data tool do not
contribute to enhancing a model.